Connect your AI or use ours
Any application and any AI can use all of your data.
Hospitals are offered new AI tools every week. The bottleneck is the data. With one unified foundation in place, each new tool connects through standard interfaces, so every introduction costs less than the one before.
AI assistant of your choice, connected via MCP
How many patients over 65 with diabetes had an HbA1c above 9% this quarter?
Calls the Firemetrics MCP server · runs a query on the data foundation
412 patients match. Most are in internal medicine (58%) and cardiology (21%). Do you want the list for the diabetes clinic?
Three ways in
Standard interfaces, the same data behind each
SQL
For IT, BI and data science teams. Query every FHIR resource with plain SQL, join across resource types, and connect the BI tools you already use.
SELECT count(DISTINCT patient_id)FROM observationWHERE code = '4548-4' -- HbA1c AND value > 8.0 AND effective > current_date - 90;FHIR API and SMART on FHIR
For applications. Standard FHIR R4 REST and SMART app launch, so any certified app can plug in. No custom interfaces, no extra interface fees.
GET /fhir/Observation ?code=http://loinc.org|4548-4 &value-quantity=gt8 &date=ge2026-07-07 &_include=Observation:patientMCP server
For AI assistants and agents. Connect Claude or any MCP-compatible tool and ask questions in plain language. It explores the schema, writes the query and shows its work.
> Which patients had an HbA1c above 8% in the last 90 days? ● firemetrics · query ↳ 214 patients · 37 new since July ↳ SQL shown · ready for a dashboardLower cost per introduction
The saving grows with every tool you add
Firemetrics follows FHIR, the global standard for health data. Regulators increasingly require it, and more and more apps and tools already speak it, so connecting the next one is far easier. On top sits a semantic layer that makes the data perfect for AI: nobody has to learn a custom schema. New hires don't need training on it, and neither does AI, because the semantic layer already explains what the data means.
Without a shared data foundation
- 1
New AI tool: own interfaces, own mapping, own clean-up.
- 2
Next app: the same data work again.
- 3
Every project pays for the data again.
With Firemetrics
- 1
Foundation once: all sources prepared in one place.
- 2
New AI tool: connects via MCP.
- 3
Next app: connects via FHIR or SQL.
Under your control
Open does not mean uncontrolled
You approve each connection
Your IT decides which application or AI tool gets access, to which data, read-only or not.
Every access is logged
Roles, permissions and an audit log for SQL, FHIR and MCP access.
No lock-in
Your data sits in open standards and can be read at any time, also if you change vendors.
access_rules
- SQLStudio · all clinical data · read-only
- FHIRPatient viewer · own ward only · read-only
- MCPAI assistant · lab results · read-only
Questions IT teams ask
Which AI tools work with the MCP server?
Any assistant or agent that supports the Model Context Protocol, once your IT has approved it.
Does patient data leave the hospital?
Not with an on-premises installation. Data only leaves your environment if you connect an external tool and approve it. FLAME, our AI assistant, runs with a local model inside your data center.
Do we need a new interface for each new application?
No. New applications connect to the same foundation via SQL or the FHIR API. There is no new integration with your source systems.
Can we use our own BI and data science tools?
Yes. Firemetrics is built on PostgreSQL, so tools like Power BI, Tableau, Python or R connect directly via SQL, or via the FHIR API.
Bring the AI tool you want to introduce next.
We show you how it connects to your data, in one session.